Showing 1 to 100 of 238 matching Articles
Results per page:
Export (CSV)
By
Adj, Gora; CervantesVázquez, Daniel; ChiDomínguez, JesúsJavier; Menezes, Alfred; RodríguezHenríquez, Francisco
Show all (5)
1 Citations
The security of the JaoDe Feo Supersingular Isogeny DiffieHellman (SIDH) key agreement scheme is based on the intractability of the Computational Supersingular Isogeny (CSSI) problem—computing
$${\mathbb F}_{p^2}$$
rational isogenies of degrees
$$2^e$$
and
$$3^e$$
between certain supersingular elliptic curves defined over
$${\mathbb F}_{p^2}$$
. The classical meetinthemiddle attack on CSSI has an expected running time of
$$O(p^{1/4})$$
, but also has
$$O(p^{1/4})$$
storage requirements. In this paper, we demonstrate that the van OorschotWiener golden collision finding algorithm has a lower cost (but higher running time) for solving CSSI, and thus should be used instead of the meetinthemiddle attack to assess the security of SIDH against classical attacks. The smaller parameter p brings significantly improved performance for SIDH.
more …
By
DelporteGallet, Carole; Fauconnier, Hugues; Rajsbaum, Sergio; Yanagisawa, Nayuta
Show all (4)
We consider a system of n anonymous processes communicating through multiwriter/multireader (MWMR) registers. A weakset object is a particularly interesting communication abstraction for anonymous processes; it may be seen as the equivalent of an atomic snapshot object in an anonymous system. It can be accessed through two operations:
$$\textsc {add}()$$
and
$$\textsc {get}()$$
. Intuitively, an
$$\textsc {add}(v)$$
operation puts value v in the set represented by the object, while a
$$\textsc {get}()$$
operation returns the contents of the set. The paper describes a waitfree atomic implementation of a weakset object shared by n anonymous processes using 3n MWMR registers. The description of the algorithm is incremental. The paper first presents an implementation that is waitfree only for the
$$\textsc {Get}()$$
operations, using 2n MWMR registers. Then it describes an implementation that is waitfree for the
$$\textsc {Get}()$$
and the
$$\textsc {Add}()$$
operations, using
$$3n+1$$
MWMR registers, and finally it is improved to an implementation using 3n MWMR registers. In addition, a lowerbound of n registers for the implementation of a waitfree atomic weakset is proved.
more …
By
Landero, V.; Pérez, Joaquín; Cruz, L.; Turrubiates, Tania; Ríos, David
Show all (5)
A review of state of art reveals that the characterization and analysis of the relation between problemalgorithm has been focused only on problem features or on algorithm features; or in some situations on both, but the algorithm logical is not considered in the analysis. The above for selecting an algorithm will give the best solution. However there is more knowledge for discovering from this relation. In this paper, significant features are proposed for describing problem structure and algorithm searching fluctuation; other known metrics were considered (Autocorrelation Coefficient and Length) but were not significant. A causal study case is performed for analyzing causes and effects from: BinPacking problem structure, Temperature, searching behavior of Threshold Accepting algorithm and final performance to solving problem instances. The proposed features permitted in the causal study to find relations causeeffect; which gave guidelines for designing a Threshold Accepting selfadaptive algorithm. Its performance outperforms to original algorithm in 74% out of 324 problem cases. The causal analysis on relevant information from problem, algorithm (both) and algorithm logical could be an important guideline to discover rules or principles over several problem domains, which permit the design of selfadaptive algorithms to give the best solution to complex problems.
more …
By
LoyolaGonzález, Octavio; MedinaPérez, Miguel Angel; MartínezTrinidad, José Fco.; CarrascoOchoa, Jesús Ariel
Show all (4)
Scientific conferences are suitable vehicles for knowledge dissemination, connecting authors, networking, and research entities. However, it is important to know the impact of a determined conference for the international research community. The main way to do this is through a scientometric study of those papers derived from the conference. Therefore, in this paper, we introduce a scientometric study taking into account all papers published in each edition of the Mexican Conference on Pattern Recognition (MCRP) as well as all the papers published in special issues derived from MCPR. Our study is based on data taken from the SCOPUS database. We have extracted and analyzed several essential keys, such as acceptance and rejection rates, number of authors and topproductive institutions, and frequency of citations by other journals, with the aim of providing the impact of the papers derived from MCPR for the international research community. From our study, we report some important findings about the impact of the MCPR conference after ten editions.
more …
By
Gutierrez, Francisco J.; Ochoa, Sergio F.; Cornejo, Raymundo; Vassileva, Julita
Show all (4)
This chapter builds upon the need to adopt a more comprehensive approach when designing computersupported technology to mediate social interaction between older adults and other generations of family members. Recognizing the complexity and heterogeneity of this communication scenario, the chapter shows the need to consider the culture as a key factor for reusing HCI design knowledge when conceiving new technology to mediate intergenerational social interactions. Considering a cultural perspective, this chapter discusses similarities and differences in the intergenerational social interaction process in Latin American and Western countries. On the one hand, the identified similitudes help reuse existing design knowledge. On the other hand, the identified differences inform the design of new solutions to mediate intergenerational communication. Understanding the underlying sociocultural traits of the social interaction scenario allows us to determine how to reuse the knowledge gained during the previous two decades of HCI research with older adults, and thus design better interaction mechanisms for the next generation of systems for this application domain.
more …
By
Marca, Yuri; Aguirre, Hernán; Martinez, Saúl Zapotecas; Liefooghe, Arnaud; Derbel, Bilel; Verel, Sébastien; Tanaka, Kiyoshi
Show all (7)
Difficult Pareto set topology refers to multiobjective problems with geometries of the Pareto set such that neighboring optimal solutions in objective space differ in several or all variables in decision space. These problems can present a tough challenge for evolutionary multiobjective algorithms to find a good approximation of the optimal Pareto set welldistributed in decision and objective space. One important challenge optimizing these problems is to keep or restore diversity in decision space. In this work, we propose a method that learns a model of the topology of the solutions in the population by performing parametric spline interpolations for all variables in decision space. We use CatmullRom parametric curves as they allow us to deal with any dimension in decision space. The proposed method is appropriated for biobjective problems since their optimal set is a onedimensional curve according to the KarushKuhnTucker condition. Here, the proposed method is used to promote restarts from solutions generated by the model. We study the effectiveness of the proposed method coupled to NSGAII and two variations of MOEA/D on problems with difficult Pareto set topology. These algorithms approach very differently the Pareto set. We argue and discuss their behavior and its implications for model building.
more …
By
LópezMonroy, A. Pastor; MontesyGómez, Manuel; Escalante, Hugo Jair; González, Fabio A.
Show all (4)
The BagofVisualWords (BoVW) representation is a well known strategy to approach many computer vision problems. The idea behind BoVW is similar to the BagofWords (BoW) used in text mining tasks: to build word histograms to represent documents. Regarding computer vision, most of the research has been devoted to obtain better visual words, rather than in improving the final representation. This is somewhat surprising, as there are many alternative ways of improving the BoW representation within the text mining community that can be applied in computer vision as well. This paper aims at evaluating the usefulness of Distributional Term Representations (DTRs) for image classification. DTRs represent instances by exploiting statistics of feature occurrences and cooccurrences along the dataset. We focus in the suitability and effectiveness of using wellknown DTRs in different image collections. Furthermore, we devise two novel distributional strategies that learn appropriated groups of images to compute better suited distributional features. We report experimental results in several image datasets showing the effectiveness of the proposed DTRs over BoVW and other methods in the literature including deep learning based strategies. In particular we show the effectiveness of the proposed representations on image collections from narrow domains, where target categories are subclasses of a more general class (e.g., subclasses of birds, aircrafts, or dogs).
more …
By
Gavrileva, U.; Alekseev, V.; Vasilyeva, M.; Basabe, J. D.; Efendiev, Y.; Gibson, R. L., Jr.
Show all (6)
In this work, we consider wave propagation in fractured media. The mathematical model is described by Helmholtz problem related to wave propagation with specific interface conditions on the fracture in the frequency domain. We use a discontinuous Galerkin method for the approximation by space that help to weakly impose interface conditions on fractures. Such approximations lead to the large system of equations and computationally expensive. In this work, we construct a coarse grid approximation for effective solution using Generalized Multiscale Discontinuous Galerkin Method (GMsDGM). In this method, we construct a multiscale space using solution of the local spectral problems in each coarse elements. The results of the numerical solution for the twodimensional problem are presented for model problems of the wave propagation in fractured media.
more …
By
Sidorov, Grigori
As we mentioned earlier in the book, in the automatic analysis of natural language (natural language processing, NLP) and in computational linguistics, machine learning methods are becoming more and more popular. Applying these methods increasingly gives better results. In this chapter, we describe the design of experiments in computational lingusitics: problem – corpus – gold standard – feature selection – dimensionality reduction – classification – evaluation (kfold cross validation).
more …
By
FalcónCardona, Jesús Guillermo; Coello Coello, Carlos A.; Emmerich, Michael
The use of multiobjective evolutionary algorithms (MOEAs) that employ a set of convex weight vectors as search directions, as a reference set or as part of a quality indicator has been widely extended. However, a recent study indicates that these MOEAs do not perform very well when tackling multiobjective optimization problem (MOPs), having different Pareto front geometries. Hence, it is necessary to propose MOEAs whose good performance is not strongly depending on certain Pareto front shapes. In this paper, we propose a Paretofront shape invariant MOEA that combines the individual effect of two indicatorbased density estimators. We selected the weakly Paretocompliant IGD
$$^+$$
indicator to promote convergence and the Riesz senergy indicator that leads to uniformly distributed point sets for the large class of rectifiable ddimensional manifolds. Our proposed approach, called CRIEMOA, is compared with respect to MOEAs that adopt convex weight vectors (NSGAIII, MOEA/D and MOMBI2) as well as to MOEAs not using this set of vectors (
$$\varDelta _p$$
MOEA and GDEMOEA) on MOPs belonging to the test suites DTLZ, DTLZ
$$^{1}$$
, WFG and WFG
$$^{1}$$
. Our experimental results show that CRIEMOA outperforms the considered MOEAs, regarding the hypervolume indicator and the SolowPolasky indicator, on most of the test problems and that its performance does not depend on the Pareto front shape of the problems.
more …
By
Sidorov, Grigori
Computational linguistics is an important area within the field of linguistics. Computational methods used in computational linguistics originate from computer science, or, to be more specific, from artificial intelligence. In fact, large part of modern computational lingusitics consists in application of machine learning methods to large textual datasets.
more …
By
Merelo, Juan J.; Laredo, J. L. J.; Castillo, Pedro A.; GarcíaValdez, JoséMario; RojasGaleano, Sergio
Show all (5)
Creating a concurrent and stateless version of an evolutionary algorithm implies changes in its algorithmic model. From the performance point of view, the main challenge is to balance computation with communication, but from the evolutionary point of view another challenge is to keep diversity high so that the algorithm is not stuck in local minima. In a concurrent setting, we will have to find the right balance so that improvements in both facets do not cancel out. In this paper we address such an issue, by exploring the space of parameters of a population based concurrent evolutionary algorithm that yields to find out the best combination for a particular problem.
more …
By
Trujillo, Leonardo; Muñoz, Luis; López, Uriel; Hernández, Daniel E.
Show all (4)
In the era of Deep Learning and Big Data, the place of Genetic Programming (GP) within the Machine Learning area seems difficult to define. Whether it is due to technical constraints or conceptual barriers, GP is currently not a paradigm of choice for the development of stateoftheart machine learning systems. Nonetheless, there are important features of the GP approach that make it unique and should continue to be actively explored and studied. In this work we focus on two aspects of GP that have previously received little or no attention, particularly in treebased GP for symbolic regression. First, on the potential of GP to perform transfer learning, where solutions evolved for one problem are transferred to another. Second, on the potential of GP individuals to detect the true underlying structure of an input dataset and detect anomalies in the input data, what are known as outliers. This work presents initial results on both issues, with the goal of fostering discussion and showing that there is still untapped potential in the GP paradigm.
more …
By
KuriMorales, Angel
A method which allows us to find a polynomial model of an unknown function from a set of tuples with n independent variables and m tuples is presented. A plausible model of an explicit approximation is found. The number of monomials of the model is determined by a previously trained neural network (NNt). The degrees of the monomials are bounded from the Universal Approximation Theorem yielding a reduced polynomial form. The coefficients of the model are found by exploring the space of possible monomials with a genetic algorithm (GA). The polynomial defined by every training pair is approximated by the Ascent Algorithm which yields the coefficients of best fit with the L_{∞} norm. The L2 error corresponding to these coefficients is calculated and becomes the fitness function of the GA. A detailed example for a well known experimental data set is presented. It is shown that the model derived from the method yields a classification error of <7% for the crossvalidation data set. A detailed description of the NNt is included. The approximation to 46 datasets tackled with our method is discussed. In all these cases the approximation accuracy was better than 94%. The tool described herein yields general models with explanatory characteristics not found in other methods.
more …
By
Castañeda, Armando; Fraigniaud, Pierre; Paz, Ami; Rajsbaum, Sergio; Roy, Matthieu; Travers, Corentin
Show all (6)
More than two decades ago, combinatorial topology was shown to be useful for analyzing distributed faulttolerant algorithms in shared memory systems and in message passing systems. In this work, we show that combinatorial topology can also be useful for analyzing distributed algorithms in networks of arbitrary structure. To illustrate this, we analyze consensus, setagreement, and approximate agreement in networks, and derive lower bounds for these problems under classical computational settings, such as the LOCAL model and dynamic networks.
more …
By
Collazos, C. A.; Rojas, Alejandra; Castellanos, Hermes; Beltrán, Diego; Collazos, César A.; Melo, Darío; Ruiz, Iván; Ostos, Iván; Sánchez, Carlos A.; DelaHozFranco, Emiro; MeléndezPertuz, Farid; Mora, César
Show all (12)
We present a quantitative analysis using the concepts of Normalized Gain and Least Squares in a process of Physics Teaching. This paper presents the results of the strategy based in The Construction of Prototypes (TCP) and Project Based Learning (PrBL) which was applied in a course of Mechanics in BogotáColombia. The strategy focuses on three topics of Rotational Dynamics Teaching (RDT) specifically at centripetal force, Inertia moment and theorem de parallel axes and angular momentum conservation. We present results and analysis of employed method.
more …
By
Cuate, Oliver; Uribe, Lourdes; Ponsich, Antonin; Lara, Adriana; Beltran, Fernanda; Sánchez, Alberto Rodríguez; Schütze, Oliver
Show all (7)
1 Citations
The recently proposed Pareto Tracer method is an effective numerical continuation technique which allows performing movements along the set of KKT points of a given multiobjective optimization problem. The nature of this predictorcorrector method leads to constructing solutions along the Pareto set/front numerically; it applies to higher dimensions and can handle box and equality constraints. We argue that the right hybridization of multiobjective evolutionary algorithms together with specific continuation methods leads to fast and reliable algorithms. Moreover, due to the continuation technique, the resulting hybrid algorithm could have a certain advantage when handling, in particular, equality constraints. In this paper, we make the first effort to hybridize NSGAII with the Pareto Tracer. To support our claims, we present some numerical results on continuously differentiable equality constrained biobjective optimization test problems, to show that the resulting hybrid NSGAII/PT is highly competitive against some stateoftheart algorithms for constrained optimization. Finally, we stress that the chosen approach could be applied to a more significant number of objectives with some adaptations of the algorithm, leading to a very promising research topic.
more …
By
LanchoBarrantes, Bárbara S.; CantúOrtiz, Francisco J.
We can find several studies analyzing the scientific production of Latin American countries, such as Argentina, Brazil, Colombia, Cuba, Guatemala, Peru, Venezuela etc throughout the scientific literature. There are many papers focusing on scientific disciplines, institutions and journals from these countries. However, to the best of our knowledge, we have not found any article that analyzes the scientific production of Mexico, global or recent and with Scopus as a specific database, nor the production in collaboration with its strategic countries in science and technology. For this reason, the present work intends to give Mexico the prominence it deserves by studying its productivity in research by using a bibliometric approach. To perform this study, the international bibliographic database Scopus was used within a ten year publication window from 2007 to 2016. With this sample, we analyzed the production in general, the scientific production in scientific disciplines, and production in collaboration with its strategic countries in science and technology, without forgetting the variables of the citations received from Scival as a parameter of impact on research. This study aims to serve as a precedent for later studies and contribute as a reference of Mexican production to the scientific community and as a tool to elaboration of national public policy in science and technology.
more …
By
Olague, Gustavo; Hernández, Daniel E.; Llamas, Paul; Clemente, Eddie; Briseño, José L.
Show all (5)
1 Citations
This work describes the use of brain programming for automating the video tracking design process. The challenge is that of creating visual programs that learn to detect a toy dinosaur from a database while tested in a visualtracking scenario. When planning an object tracking system, two subtasks need to be approached: detection of moving objects in each frame and correct association of detection to the same object over time. Visual attention is a skill performed by the brain whose functionality is to perceive salient visual features. The automatic design of visual attention programs through an optimization paradigm is applied to the detectionbased tracking of objects in a video from a moving camera. A system based on the acquisition and integration steps of the natural dorsal stream was engineered to emulate its selectivity and goaldriven behavior useful to the task of tracking objects. This is considered a challenging problem since many difficulties can arise due to abrupt object motion, changing appearance patterns of both the object and the scene, nonrigid structures, objecttoobject and objecttoscene occlusions, as well as camera motion, models, and parameters. Tracking relies on the quality of the detection process and automatically designing such stage could significantly improve tracking methods. Experimental results confirm the validity of our approach using three different kinds of robotic systems. Moreover, a comparison with the method of regions with convolutional neural networks is provided to illustrate the benefit of the approach.
more …
By
Neri Mendoza, Verónica; Ledeneva, Yulia; GarcíaHernández, René Arnulfo
MultiDocument Text Summarization (MDTS) consists of generating an abstract from a group of two or more number of documents that represent only the most important information of all documents. Generally, the objective is to obtain the main idea of several documents on the same topic. In this paper, we propose a new MDTS method based on a Genetic Algorithm (GA). The fitness function is calculated considering two text features: sentence position and coverage. We propose the binary coding representation, selection, crossover and mutation operators to improve the stateoftheart results. We test the proposed method on DUC02 data set, specifically, on Abstractive MultiDocument Text Summarization (AMDST) task demonstrating the improvement over the stateofart methods. Four different tasks for each of the 59 collection of documents (in total 567 documents) are tested. In addition, we test different configurations of the most used methodology to generate AMDST summaries. Moreover, different heuristics such as topline, baseline, baselinerandom and lead baseline are calculated. The proposed method for AMDTS demonstrates the improvement over the stateofart methods and heuristics.
more …
By
Cuate, Oliver; Schütze, Oliver
In almost all cases the performance of a multiobjective evolutionary algorithm (MOEA) is measured in terms of its approximation quality in objective space. As a consequence, most MOEAs focus on such approximations while neglecting the distribution of the individuals in decision space. This, however, represents a potential shortcoming in certain applications as in many cases one can obtain the same or a very similar qualities (measured in objective space) in several ways (measured in decision space) which may be very valuable information for the decision maker for the realization of a project.
In this work, we propose the variableNSGAIII (vNSGAIII) an algorithm that performs an exploration both in objective and decision space. The idea behind this algorithm is the socalled variation rate, a heuristic that can easily be integrated into other MOEAs as it is free of additional design parameters. We demonstrate the effectiveness of our approach on several benchmark problems, where we show that, compared to other methods, we significantly improve the approximation quality in decision space without any loss in the quality in objective space.
more …
By
Uribe, Lourdes; Schütze, Oliver; Lara, Adriana
Multiobjective optimization problems (MOPs) arise in a natural way in diverse knowledge areas. Multiobjective evolutionary algorithms (MOEAs) have been applied successfully to solve this type of optimization problems over the last two decades. However, until now MOEAs need quite a few resources in order to obtain acceptable Pareto set/front approximations. Even more, in certain cases when the search space is highly constrained, MOEAs may have troubles when approximating the solution set. When dealing with constrained MOPs (CMOPs), MOEAs usually apply penalization methods. One possibility to overcome these situations is the hybridization of MOEAs with local search operators. If the local search operator is based on classical mathematical programming, gradient information is used, leading to a relatively high computational cost. In this work, we give an overview of our recently proposed constraint handling methods and their corresponding hybrid algorithms. These methods have specific mechanisms that deal with the constraints in a wiser way without increasing their cost. Both methods do not explicitly compute the gradients but extract this information in the best manner out of the current population of the MOEAs. We conjecture that these techniques will allow for the fast and reliable treatment of CMOPs in the near future. Numerical results indicate that these ideas already yield competitive results in many cases.
more …
By
LazoCortés, Manuel S.; MartínezTrinidad, José Fco.; CarrascoOchoa, Jesús A.
In Rough Set Theory, superreducts are subsets of attributes that retain the ability of the whole set of attributes to discern objects belonging to different classes; reducts are minimal ones. On the other hand, constructs also allow discerning objects belonging to different classes but, at the same time, they retain similarities between objects belonging to the same. Therefore, constructs are a kind of superreducts in whose definition interclass and intraclass information is combined. This type of superreduct has been little studied. In this paper, we present a case study, about the use of constructs instead of reducts for building decision rules useful for rulebased classification. Our results show the practical utility of constructs for rule based classification.
more …
By
CéspedesHernández, David; GonzálezCalleros, Juan Manuel; GuerreroGarcía, Josefina; RodríguezVizzuett, Liliana
Show all (4)
Natural user interaction has been recently identified as a field of interest of computer sciences, mainly due to the development of hardware devices that enable fast identification and recognition of gestures within a diversity of contexts. In this paper, the investigation of body gestures that users naturally perform to navigate virtual reality environments is reported. For this purpose, users were asked to show the commands they would perform when realizing basic navigation tasks within a virtual world, and they were observed following the Wizard of Oz approach. The performance gesturebased interaction is evaluated in terms of usability, contrasted with traditional desktop interaction, and analyzed. Finally, as result of this work, a set of gestures (language) to navigate virtual reality environments is defined, along with insights regarding this interaction modality, and comments on the future direction of natural user interfaces using fullbody movement as input.
more …
By
Camacho, Auraham; Toscano, Gregorio; Landa, Ricardo; Ishibuchi, Hisao
Show all (4)
Weight adaptation methods can enhance the diversity of solutions obtained by decompositionbased approaches when addressing irregular Pareto front shapes. Generally, these methods adapt the location of each weight vector during the search process. However, early adaptation could be unnecessary and ineffective because the population does not provide a good Pareto front approximation at early generations. In order to improve its performance, a better approach would be to trigger such adaptation only when the population has reached the Pareto front. In this paper, we introduce a performance indicator to assist weight adaptation methods, called the median of dispersion of the population (MDP). The proposed indicator provides a general snapshot of the progress of the population toward the Pareto front by analyzing the local progress of each subproblem. When the population becomes steady according to the proposed indicator, the adaptation of weight vectors starts. We evaluate the performance of the proposed approach in both regular and irregular test problems. Our experimental results show that the proposed approach triggers the weight adaptation when it is needed.
more …
By
Fausto, Fernando; ReynaOrta, Adolfo; Cuevas, Erik; Andrade, Ángel G.; PerezCisneros, Marco
Show all (5)
2 Citations
Natureinspired metaheuristics comprise a compelling family of optimization techniques. These algorithms are designed with the idea of emulating some kind natural phenomena (such as the theory of evolution, the collective behavior of groups of animals, the laws of physics or the behavior and lifestyle of human beings) and applying them to solve complex problems. Natureinspired methods have taken the area of mathematical optimization by storm. Only in the last few years, literature related to the development of this kind of techniques and their applications has experienced an unprecedented increase, with hundreds of new papers being published every single year. In this paper, we analyze some of the most popular natureinspired optimization methods currently reported on the literature, while also discussing their applications for solving realworld problems and their impact on the current literature. Furthermore, we open discussion on several research gaps and areas of opportunity that are yet to be explored within this promising area of science.
more …
By
Escobar, Carlos A.; Wegner, Diana M.; Gaur, Abhinav; MoralesMenendez, Ruben
Show all (4)
Process Monitoring for Quality is a manufacturing quality philosophy aimed at defect detection through binary classification that is founded on big data and big models. Genetic Programming (GP) algorithms have been successfully applied by following the big models learning paradigm for rare quality event detection (classification). Since it is a biasfree technique unmarred by human preconceptions, it can potentially generate better solutions (models) compared with the best human efforts. However, since GP uses random search methods based on Darwinian philosophy of “survival of the fittest”, hundreds, or even thousands of models need to be created to find a good solution. In this context, model selection becomes a critical step in the process of finding the final model to be deployed at the plant. A threeobjective optimization model selection criterion (
$$3DGP$$
) is introduced for analyzing highly/ultra unbalanced data structures. It uses three competing attributes – prediction, separability, complexity – to project candidate models into a threedimensional space to select the final model that solves the posed tradeoff between them the best.
more …
By
Yu, Wen; Li, Xiaoou; Gonzalez, Jesus
Deep recurrent neural networks (RNN), such as LSTM, have many advantages over forward networks. However, the LSTM training method, such as backward propagation through time (BPTT), is really slow.
In this paper, by separating the LSTM cell into forward and recurrent substructures, we propose a much simpler and faster training method than the BPTT. The deep LSTM is modified by combining the deep RNN with the multilayer perceptron (MLP). The simulation results show that our fast training method for LSTM is better than BPTT for LSTM.
more …
By
Wong, Hallee E.; Akar, Osman; Cuevas, Emmanuel Antonio; Tabian, Iuliana; Ravichandran, Divyaa; Fu, Iris; Carter, Cambron
Show all (7)
Markerless augmented reality can be a challenging computer vision task, especially in live broadcast settings and in the absence of information related to the video capture such as the intrinsic camera parameters. This typically requires the assistance of a skilled artist, along with the use of advanced video editing tools in a postproduction environment. We present an automated video augmentation pipeline that identifies textures of interest and overlays an advertisement onto these regions. We constrain the advertisement to be placed in a way that is aesthetic and natural. The aim is to augment the scene such that there is no longer a need for commercial breaks. In order to achieve seamless integration of the advertisement with the original video we build a 3D representation of the scene, place the advertisement in 3D, and then project it back onto the image plane. After successful placement in a single frame, we use homographybased, shapepreserving tracking such that the advertisement appears perspective correct for the duration of a video clip. The tracker is designed to handle smooth camera motion and shot boundaries.
more …
By
Takato, Setsuo; Vallejo, José A.
Hamiltonian dynamical systems can be studied from a variety of viewpoints. Our intention in this paper is to show some examples of usage of two Maxima packages for symbolical and numerical analysis (pdynamics and poincare, respectively), along with the set of scripts
for obtaining the
code corresponding to graphical representations of Poincaré sections, including animation movies.
more …
By
Chairez, Isaac; Poznyak, Alexander; Nazin, Alexander; Poznyak, Tatyana
Show all (4)
A new method to design learning laws for neural networks with continuous dynamics is proposed in this study. The learning method is based on the socalled doubleaveraged descendant technique (DASGDT), which is a variant of the gradientdescendant method. The learning law implements a double averaged algorithm which filters the effect of uncertainties of the states, which are continuously measurable. The learning law overcomes the classical assumption on the strict convexity of the functional with respect to the weights. The photocatalytic ozonation process of a single contaminant is estimated using the learning law design proposed in this study.
more …
By
HernándezRubio, Erika; MenesesViveros, Amilcar; Muñoz Salazar, Laura
In Mexico, the health gap has increased, so that the population with health problems exceeds the capacity of the available medical specialists. The population sector of the elderly has difficulty moving to medical care sites to undergo treatments. The use of eHealthy tools can help reduce the problem of the health gap by expanding the coverage of medical care to sectors of the population that have difficulties accessing health services. Neuropsychological tests can be digitized on mobile devices and help in the area of neuropsychology. It has been detected that tablets are an ideal mobile device for older adults due to the size of their screen and the different types of interaction they offer. Several neuropsychological tests have been developed: 10word learning, Poppelreuter and Raven, among others, in tablets. In this paper we present the results of user experiences when testing these applications in seniors in various centers for older adults in Mexico City.
more …
By
SerranoRubio, J. P.; Everson, Richard
This paper presents an automatic method for brain tumour segmentation from magnetic resonance images. The method uses the feature vectors obtained by an efficient feature encoding approach which combines the advantages of the supervoxels and sparse coding techniques. Extremely Randomized Trees (ERT) algorithm is trained using these feature vectors to detect the whole tumour and for multilabel classification of abnormal tissues. A Conditional Random Field (CRF) algorithm is implemented to delimit the region where the brain tumour is located. The obtained predictions of the ERT are used to estimate probability maps. The probability maps of the images and the Euclidean distance between the feature vectors of neighbour supervoxels define the conditional random field energy function. The minimization of the energy function is performed via graph cuts. The proposed methods are evaluated on real patient data obtained from BraTS 2018 challenge. Results demonstrate that proposed method achieves a competitive performance on the validation dataset using Dice score is: 0.5719, 0.7992 and 0.6285 for enhancing tumuor, whole tumour and tumour core respectively. The achieved performance of this method on testing set using Dice score is: 0.5081, 0.7278 and 0.5778 for enhancing tumuor, whole tumour and tumour core respectively.
more …
By
Torres, Leo; SuárezSerrato, Pablo; EliassiRad, Tina
Graph distance and graph embedding are two fundamental tasks in graph mining. For graph distance, determining the structural dissimilarity between networks is an illdefined problem, as there is no canonical way to compare two networks. Indeed, many of the existing approaches for network comparison differ in their heuristics, efficiency, interpretability, and theoretical soundness. Thus, having a notion of distance that is built on theoretically robust first principles and that is interpretable with respect to features ubiquitous in complex networks would allow for a meaningful comparison between different networks. For graph embedding, many of the popular methods are stochastic and depend on blackbox models such as deep networks. Regardless of their high performance, this makes their results difficult to analyze which hinders their usefulness in the development of a coherent theory of complex networks. Here we rely on the theory of the length spectrum function from algebraic topology, and its relationship to the nonbacktracking cycles of a graph, in order to introduce two new techniques: NonBacktracking Spectral Distance (NBD) for measuring the distance between undirected, unweighted graphs, and NonBacktracking Embedding Dimensions (NBED) for finding a graph embedding in lowdimensional space. Both techniques are interpretable in terms of features of complex networks such as presence of hubs, triangles, and communities. We showcase the ability of NBD to discriminate between networks in both real and synthetic data sets, as well as the potential of NBED to perform anomaly detection. By taking a topological interpretation of nonbacktracking cycles, this work presents a novel application of topological data analysis to the study of complex networks.
more …
By
Tchernykh, Andrei; MirandaLópez, Vanessa; Babenko, Mikhail; ArmentaCano, Fermin; Radchenko, Gleb; Drozdov, Alexander Yu.; Avetisyan, Arutyun
Show all (7)
2 Citations
Properties of redundant residue number system (RRNS) are used for detecting and correcting errors during the data storing, processing and transmission. However, detection and correction of a single error require significant decoding time due to the iterative calculations needed to locate the error. In this paper, we provide a performance evaluation of AsmuthBloom and Mignotte secret sharing schemes with three different mechanisms for error detecting and correcting: Projection, Syndrome, and ARRRNS. We consider the best scenario when no error occurs and worstcase scenario, when error detection needs the longest time. When examining the overall coding/decoding performance based on real data, we show that ARRRNS method outperforms Projection and Syndrome by 68% and 52% in the worstcase scenario.
more …
By
Pellegrin, Luis; Escalante, Hugo Jair; MontesyGómez, Manuel; González, Fabio A.
Show all (4)
Automatic Image Annotation (AIA) is the task of assigning keywords to images, with the aim to describe their visual content. Recently, an unsupervised approach has been used to tackle this task. Unsupervised AIA (UAIA) methods use reference collections that consist of the textual documents containing images. The aim of the UAIA methods is to extract words from the reference collection to be assigned to images. In this regard, by using an unsupervised approach it is possible to include large vocabularies because any word could be extracted from the reference collection. However, having a greater diversity of words for labeling entails to deal with a larger number of wrong annotations, due to the increasing difficulty for assigning a correct relevance to the labels. With this problem in mind, this paper presents a general strategy for UAIA methods that reranks assigned labels. The proposed method exploits the semanticrelatedness information among labels in order to assign them an appropriate relevance for describing images. Experimental results in different benchmark datasets show the flexibility of our method to deal with assignments from freevocabularies, and its effectiveness to improve the initial annotation performance for different UAIA methods. Moreover, we found that (1) when considering the semanticrelatedness information among the assigned labels, the initial ranking provided by a UAIA method is improved in most of the cases; and (2) the robustness of the proposed method to be applied on different UAIA methods, will allow extending capabilities of stateoftheart UAIA methods.
more …
By
Carvalho, Cícero; RamírezMondragón, Xavier; Neumann, Victor G. L.; TapiaRecillas, Horacio
Show all (4)
In 1988 Lachaud introduced the class of projective Reed–Muller codes, defined by evaluating the space of homogeneous polynomials of a fixed degree d on the points of
$$\mathbb {P}^n(\mathbb {F}_q)$$
. In this paper we evaluate the same space of polynomials on the points of a higher dimensional scroll, defined from a set of rational normal curves contained in complementary linear subspaces of a projective space. We determine a formula for the dimension of the codes, and the exact value of the dimension and the minimum distance in some special cases.
more …
By
ValenciaRosado, Luis Oswaldo; Starostenko, Oleg
Advances in computer graphics allow to simulate ever growing virtual worlds with a higher level of realism which can even be created in real time. An integral part of these worlds are the terrains which are the physical features of the land. Despite the capabilities of modern computer systems, the creation process still demands high amounts of manhours. To automatically generate coherent, realisticlooking and useful content is still and open problem, and research focuses on how to automatize these processes while allowing users to exert a certain degree on control over the generated content. This survey goes over the different techniques used for the automatic generation of terrains, which include different land formations such as mountains, valleys, rivers, shores, etc. These terrains have different uses such as simulation or entertainment which translates on different needs over the desired realism of the terrain and the degree of control that users have. Through time, different approaches have been proposed: repeating patterns that resemble those seen in nature; using software agents that imitate geological processes; using artificial intelligence techniques for pattern recognition and imitation of landscapes; or allowing users to interact with the system to draw desired terrain features. This review presents an overview of the area and discusses how different techniques adapt to the different needs and different stages of terrain creation.
more …
By
Sidorov, Grigori
Another idea related to the nonlinear construction of ngrams, i.e., using distinct elements or distinct order of their appearance in a text, is the idea of replacing words by their synonyms or by the generalized concepts that correspond to the words according to a certain ontology.
more …
By
Ramírez, Gabriela; Villatoro, Esaú; Ionescu, Bogdan; Escalante, Hugo Jair; Escalera, Sergio; Larson, Martha; Müller, Henning; Guyon, Isabelle
Show all (8)
Progress in the autonomous analysis of human behavior from multimodal information has lead to very effective methods able to deal with problems like action/gesture/activity recognition, pose estimation, opinion mining, user tailored retrieval, etc. However, it is only recently that the community has been starting to look into related problems associated with more complex behavior, including personality analysis, deception detection, among others. We organized an academic contest colocated with ICPR2018 running two tasks in this direction. On the one hand, we organized an information fusion task in the context of multimodal image retrieval in social media. On the other hand, we ran another task in which we aim to infer personality traits from written essays, including textual and handwritten information. This paper describes both tasks, detailing for each of them the associated problem, data sets, evaluation metrics and protocol, as well as an analysis of the performance of simple baselines.
more …
By
Raza, Mushtaq; Faria, João Pascoal; Salazar, Rafael
Collecting product and process measures in software development projects, particularly in education and training environments, is important as a basis for assessing current performance and opportunities for improvement. However, analyzing the collected data manually is challenging because of the expertise required, the lack of benchmarks for comparison, the amount of data to analyze, and the time required to do the analysis. ProcessPAIR is a novel tool for automated performance analysis and improvement recommendation; based on a performance model calibrated from the performance data of many developers, it automatically identifies and ranks potential performance problems and root causes of individual developers. In education and training environments, it increases students’ autonomy and reduces instructors’ effort in grading and feedback. In this article, we present the results of a controlled experiment involving 61 software engineering master students, half of whom used ProcessPAIR in a Personal Software Process (PSP) performance analysis assignment, and the other half used a traditional PSP support tool (Process Dashboard) for performing the same assignment. The results show significant benefits in terms of students’ satisfaction (average score of 4.78 in a 1–5 scale for ProcessPAIR users, against 3.81 for Process Dashboard users), quality of the analysis outcomes (average grades achieved of 88.1 in a 0–100 scale for ProcessPAIR users, against 82.5 for Process Dashboard users), and time required to do the analysis (average of 252 min for ProcessPAIR users, against 262 min for Process Dashboard users, but with much room for improvement).
more …
By
Muraña, Jonathan; Nesmachnow, Sergio; Armenta, Fermín; Tchernykh, Andrei
Show all (4)
This article presents an empirical evaluation of power consumption for scientific computing applications in multicore systems. Three types of applications are studied, in single and combined executions on Intel and AMD servers, for evaluating the overall power consumption of each application. The main results indicate that power consumption behavior has a strong dependency with the type of application. Additional performance analysis shows that the best load of the server regarding energy efficiency depends on the type of the applications, with efficiency decreasing in heavily loaded situations. These results allow formulating a model to characterize applications according to power consumption, efficiency, and resource sharing, which provide useful information for resource management and scheduling policies. Several scheduling strategies are evaluated using the proposed energy model over realistic scientific computing workloads. Results confirm that strategies that maximize host utilization provide the best energy efficiency and performance results.
more …
By
Alcántara, Manuel; Castañeda, Armando; FloresPeñaloza, David; Rajsbaum, Sergio
Show all (4)
LookComputeMove models for a set of autonomous robots have been thoroughly studied for over two decades. We consider the standard Asynchronous Luminous Robots (ALR) model, where robots are located in a graph G. Each robot, repeatedly Looks at its surroundings and obtains a snapshot containing the vertices of G, where all robots are located; based on this snapshot, each robot Computes a vertex (adjacent to its current position), and then Moves to it. Robots have visible lights, allowing them to communicate more information than only its actual position, and they move asynchronously, meaning that each one runs at its own arbitrary speed. We are also interested in a case which has been barely explored: the robots need not all be present initially, they might appear asynchronously. We call this the Extended Asynchronous Appearing Luminous Robots (EALR) model. A central problem in the mobile robots area is bringing the robots to the same vertex. We study several versions of this problem, where the robots move towards the same (or close to each other) vertices. And we concentrate on the requirement that each robot executes a finite number of LookComputeMove cycles, independently of the interleaving of other robot’s cycles, and then stops. Our main result is direct connections between the (ALR and) EALR model and the asynchronous waitfree multiprocess read/write shared memory (WFSM) model. General robot tasks in a graph are also provided, which include several version of gathering. Finally, using the connection between the EALR model and the WFSM model, a combinatorial topology characterization for the solvable robot tasks is presented.
more …
By
Ruiz, Iván; Collazos, C. A.; Sanz, Fredy A.; García, José; DelaHozFranco, Emiro; MeléndezPertuz, Farid; Mora, César
Show all (7)
Several technologies in electronics and informatics have been merged into medicine field. It is commonly called telemedicine. Specifically, this work proposes to scenario and an environment which a specific protocol of physical rehabilitation process can be executed and applied to a patient by a healthcare specialist who are geographically distant one another. In this sense, a Service Oriented Architecture (SOA) was designed and analyzed, trying to ensure the right communication process. Colored Petri Nets was used as a formal modelling tool. Finally, a brief mathematical approach is shown aiming to validate the architecture specifications.
more …
By
ValdezRodríguez, José E.; Calvo, Hiram; FelipeRiverón, Edgardo M.
The task consists in estimating the personality traits of users from their handwritten texts. To classify them, we use the scanned image of the subject handwritten essay divided in patches and we propose in this work an architecture based on a Convolutional Neural Network (CNN) as classifier. The original dataset consists of 418 images in color, from which we obtained 216 patches of each image in grayscale and then we binarized them resulting in approximately 90,000 images. The CNN consists of five convolutional layers to extract features of the patches and three fully connected layers to perform the classification.
more …
By
Sidorov, Grigori
The vector space model is a widely used model in computer science. Its wide use is due to the simplicity of the model and its very clear conceptual basis that corresponds to the human intuition in processing information and data. The idea behind the model is very simple, and it is an answer to the question, how can we compare objects in a formal way? It seems that the only way to describe the objects is to use a representation with features (characteristics) and their values. It is a universal idea, and it even seems to be the only possible way to work with formal objects.
more …
By
Pérez y Pérez, Rafael; Guerrero Román, Iván
This paper describes a computer agent for the automatic generation of visual compositions based on the EngagementReflection Model of creative writing (Pérez y Pérez Cogn. Syst. Res. 8, 89–109, 2007; Pérez y Pérez and Sharples J. Exp. Theor. Artif. Intell. 13, 119–139, 2001). During engagement the system progresses the composition; during reflection the agent evaluates, and if necessary modifies, the material produced so far and generates a set of guidelines that constrains the production of material during engagement. The final output is the result of a constant interplay between these two states. We offer details of the model and describe a prototype that provides the users with the possibility of adding compositions to the knowledgebase. Then, we show how through engagement and reflection cycles, the system is capable of generating novel outputs. Using a questionnaire, we asked a group of volunteers to describe the features of pieces produced by the program and the features of pieces produced by human designers. The results suggest that our agent provides an adequate novel framework to study the generation of automatic visual compositions.
more …
By
Guyon, Isabelle; SunHosoya, Lisheng; Boullé, Marc; Escalante, Hugo Jair; Escalera, Sergio; Liu, Zhengying; Jajetic, Damir; Ray, Bisakha; Saeed, Mehreen; Sebag, Michèle; Statnikov, Alexander; Tu, WeiWei; Viegas, Evelyne
Show all (13)
The ChaLearn AutoML Challenge (The authors are in alphabetical order of last name, except the first author who did most of the writing and the second author who produced most of the numerical analyses and plots.) (NIPS 2015 – ICML 2016) consisted of six rounds of a machine learning competition of progressive difficulty, subject to limited computational resources. It was followed bya oneround AutoML challenge (PAKDD 2018). The AutoML setting differs from former model selection/hyperparameter selection challenges, such as the one we previously organized for NIPS 2006: the participants aim to develop fully automated and computationally efficient systems, capable of being trained and tested without human intervention, with code submission. This chapter analyzes the results of these competitions and provides details about the datasets, which were not revealed to the participants. The solutions of the winners are systematically benchmarked over all datasets of all rounds and compared with canonical machine learning algorithms available in scikitlearn. All materials discussed in this chapter (data and code) have been made publicly available at
http://automl.chalearn.org/
.
more …
By
Virgilio G, Carlos D.; Sossa, Humberto; Antelis, Javier M.; Falcón, Luis E.
Show all (4)
We report the development and evaluation of brain signal classifiers, specifically Spiking Neuron based classifiers. The proposal consists of two main stages: feature extraction and pattern classification. The EEG signals used represent four motor imagery tasks: Left Hand, Right Hand, Foot and Tongue movements. In addition, one more class was added: Rest. These EEG signals were obtained from a database provided by the Technological University of Graz. Feature extraction stage was carried out by applying two algorithms: Power Spectral Density and Wavelet Decomposition. The tested algorithms were: KNearest Neighbors, Multilayer Perceptron, Single Spiking Neuron and Spiking Neural Network. All of them were evaluated in the classification between two Motor Imagery tasks; all possible pairings were made with the 5 mental tasks (Rest, Left Hand, Right Hand, Tongue and Foot). In the end, a performance comparison was made between a Multilayer Perceptron and Spiking Neural Network.
more …
By
Flores, Susana; Torrero, Claudia; Torrero, Everardo; Handam, Lamia; Flores, Silvana
Show all (5)
The change of strategies in teaching in Higher Education motivated the present work. With the application of cooperative and collaborative learning in the classroom, better results were obtained in the subject Research Workshop II, which is taught in Higher Education in Engineering. The work applying cooperative and collaborative learning impacted on the way to evaluate the participants, generating the need to find a way that allowed to have the elements to know who was contributing, which team progressed slower according to the program, as well as their peers in the development of the research project. In recent years in a significant number of companies dedicated to Information Technology and Communications have incorporated agile methodologies to manage different projects in very diverse branches, but especially in Software Engineering. In particular, the methodology used is Scrum. Therefore, Scrum was incorporated to develop students’ projects, but also to evaluate the progress that students are making in the development of projects. With the incorporation of Scrum, better results have been obtained in the projects that the students develop, they and the evaluator are aware of the progress and delays they have in achieving them, in order to take actions that allow the products requested to be completed in a timely manner the curricular program of the subject.
more …
By
Sidorov, Grigori
So, we have already learned how to obtain syntactic ngrams (although, at the moment, we are considering only continuous syntactic ngrams). Now let’s discuss what types of syntactic ngrams exist depending on the elements they are formed of, i.e., what kind of elements (components) can be parts of syntactic ngrams. In fact, the considerations to be discussed are the same for any type of ngrams.
more …
By
Sidorov, Grigori
In the previous chapters, we introduced the new concept of syntactic ngrams, i.e., ngrams obtained following paths in syntax trees. The discussion that follows in this chapter addresses the comparison of continuous syntactic ngrams with noncontinuous syntactic ngrams (i.e., ngrams with bifurcations (ramifications) and ngrams without them).
more …
By
JiménezCabas, Javier; MeléndezPertuz, Farid; OvallosGazabon, David; VélezZapata, Jaime; Castellanos, Hermes E.; Cárdenas, César A.; Sánchez, Joaquín F.; Jimenez, Gonzalo; Mora, César; Sanz, Fredy A.; Collazos, C. A.
Show all (11)
Evaporation is a process that is widely used in the chemical industry and aims to concentrate a solution consisting of a nonvolatile solute and a volatile solvent. In this paper the design of robust control systems for a simple effect evaporation system is presented. Two controllers were designed, the first was based on the Algebraic Riccati Equations (ARE) solutions technique and the second was derived from the DK iteration method. To show the potentiality of the control system proposed, we present the results of some tests carried out in simulation.
more …
By
RodriguezCoayahuitl, Lino; MoralesReyes, Alicia; Escalante, Hugo Jair
In recent years Convolutional Neural Networks (CNN) have come to dominate many machine learning tasks, specially those related to image analysis, such as object recognition. Herein we explore the possibility of developing image denoising filters by stacking multiple Genetic Programming (GP) syntax trees, in a similar fashion to how CNNs are designed. We test the evolved filters performance in removing additive Gaussian noise. Results show that GP is able to generate a diverse set of feature maps at the ’hidden’ layers of the proposed architecture. Although more research is required to validate the suitability of GP for image denoising, our work set the basis for bridging the gap between deep learning and evolutionary computation.
more …
By
Costa, Ernesto Pérez; VillaseñorPienda, Luis; Morales, Eduardo; Escalante, Hugo Jair
Show all (4)
Personality has been considered as one of the most difficult human attributes to understand. It is very important as it can be used to define the uniqueness of a person, and it has a direct impact into several aspects of everyone’s life. This paper describes our participation in the HWxPI challenge @ ICPR 2018, an academic competition focusing on the development of methods for estimation of apparent personality from handwritten and textual information. The proposed solution combined information extracted from both text and images. From the textual modality, words, and other linguistic features were considered; whereas handwritten information was represented with shape features extracted from segmented characters. Although the task was extremely hard, we show that the considered features indeed convey useful information that can be used to estimate apparent personality.
more …
By
OrtizBarrios, Miguel; Pancardo, Pablo; JiménezDelgado, Genett; ÁvilaVillalobos, Jeferson
Show all (4)
Accident and Emergency departments (A&ED) are in charge of providing access to patients requiring urgent acute care. A&ED are difficult to model due to the presence of interactions, different pathways and the multiple outcomes that patients may undertake depending on their health status. In addition, public concern has focused on the presence of overcrowding, long waiting times, patient dissatisfaction and cost overruns associated with A&ED. There is then a need for tackling these problems through developing integrated and explicit models supporting healthcare planning. However, the studies directly concentrating on modelling the A&EDs are largely limited. Therefore, this paper presents the use of a multiphase DES framework for modelling the A&ED and facilitating the assessment of potential improvement strategies. Initially, the main components, critical variables and different states of the A&ED are identified to correctly model the entire patient journey. In this step, it is also necessary to characterize the demand in order to categorize the patients into pipelines. After this, a discreteevent simulation (DES) model is developed. Then, validation is conducted through the 2sample t test to demonstrate whether the model is statistically comparable with the realworld A&ED department. This is followed by the use of Markov phasetype models for calculating the total costs of the whole system. Finally, various scenarios are explored to assess their potential impact on multiple outcomes of interest. A case study of a mixedpatient environment in a private A&E department is provided to validate the effectiveness of the multiphase DES approach.
more …
By
Furlán, Federico; Rubio, Elsa; Sossa, Humberto; Ponce, Víctor
Show all (4)
In this paper we study the problem of rock detection in a Marslike environment. We propose a convolutional neural network (CNN) to obtain a segmented image. The CNN is a modified version of the Unet architecture with a smaller number of parameters to improve the inference time. The performance of the methodology is proved in a dataset that contains several images of a Marslike environment, achieving an Fscore of 78.5%.
more …
By
Arce, Fernando; MendozaMontoya, Omar; Zamora, Erik; Antelis, Javier M.; Sossa, Humberto; CantilloNegrete, Jessica; CarinoEscobar, Ruben I.; Hernández, Luis G.; Falcón, Luis Eduardo
Show all (9)
Dendrite ellipsoidal neurons are a novel and different alternative for classification tasks, giving competitive results compared with typical classification methods. Based on kmeans++ algorithm, the network allows each dendrite to build a hyperellipsoidal in order to assign each incoming pattern
$$x_{i}=(x_{1},x_{2},\ldots ,x_{n})^{T}$$
to its respective C class. The main disadvantage of this training algorithm is the lack of accuracy in high dimensional datasets. In this research, we solved this problem by training the dendrite ellipsoidal neuron using stochastic gradient descent. Furthermore, electroencephalography data were acquired during two mental conditions (imaginary movements of the left and right hand) in order to test the new training algorithm. The proposed algorithm outperformed the accuracy acquired by a dendrite ellipsoidal neuron based on kmeans++ obtaining 76.02% and 62.77%, respectively. Also, the algorithm was compared with multilayer perceptrons and support vector machines which are some of the most common classifiers used to detect motorrelated information in brain signals. These achieved an accuracy of 72.38% and 65.81%, respectively.
more …
By
Sidorov, Grigori
The question arises, how to represent noncontinuous syntactic ngrams without resorting to their graphic form? Recall that continuous syntactic ngrams are simply sequences of words (obtained by following paths in a syntactic tree), but the case of the noncontinuous syntactic ngram is rather different.
more …
By
Calle, Laura; Merelo, Juan J.; MoraGarcía, Antonio; GarcíaValdez, JoséMario
Show all (4)
This paper presents an original approach for building structures that are stable under gravity for the physicsbased puzzle game Angry Birds, with the ultimate objective of creating levels with the minimum number of constraints. This approach consists of a searchbased procedural level generation method that uses evolutionary algorithms. In order to evaluate the stability of the levels, they are executed in an adaptation of an open source version of the game called Science Birds. In the same way, an open source evolutionary computation framework has been implemented to fit the requirements of the problem. The main challenge has been to design a fitness function that, first, avoids if possible the actual execution of the simulator, which is time consuming, and, then, to take into account the different ways in which a structure is not structurally sound and consider them in different ways to provide a smooth landscape that eventually achieves that soundness. Different representations and operators have been considered and studied. In order to test the method four experiments have been carried out, obtaining a variety of stable structures, which is the first path for the generation of levels that are aesthetically pleasing as well as playable.
more …
By
RamírezDíaz, Adrián J.; RodríguezGarcía, José; Mendoza, Sonia; Viveros, Amilcar Meneses
Show all (4)
In ubiquitous computing systems, determining the location of objects in the environment can provide basic information about the context of such objects. In closed environments an Interior Positioning System (IPS) helps to determine the location of people or robots through the use of a pointbased reference system placed in the environment. Several mechanisms can be used to locate references, for example: light sensing, radio frequencies, sound, or images. In this paper, it is presented an imagebased IPS that finds the location of a robot in a zone and provides functions to generate paths for the robot. The zones are identified through reference markers, which are analyzed in a server using image processing and Cloud Robotics, in order to minimize processing load in the robot. Once the marker is analyzed, a route is sent to the robot.
more …
By
CastilloRamirez, Alonso; SanchezAlvarez, Miguel
For a group G and a finite set A, denote by
$$\mathrm {CA}(G;A)$$
the monoid of all cellular automata over
$$A^G$$
and by
$$\mathrm {ICA}(G;A)$$
its group of units. We study the minimal cardinality of a generating set, known as the rank, of
$$\mathrm {ICA}(G;A)$$
. In the first part, when G is a finite group, we give upper bounds for the rank in terms of the number of conjugacy classes of subgroups of G. The case when G is a finite cyclic group has been studied before, so here we focus on the cases when G is a finite dihedral group or a finite Dedekind group. In the second part, we find a basic lower bound for the rank of
$$\mathrm {ICA}(G;A)$$
when G is a finite group, and we apply this to show that, for any infinite abelian group H, the monoid
$$\mathrm {CA}(H;A)$$
is not finitely generated. The same is true for various kinds of infinite groups, so we ask if there exists an infinite group H such that
$$\mathrm {CA}(H;A)$$
is finitely generated.
more …
By
MartinezRodriguez, Jose L.; LopezArevalo, Ivan; RiosAlvarado, Ana B.; Hernandez, Julio; AldanaBobadilla, Edwin
Show all (5)
The vision of the Semantic Web is to get information with a defined meaning in a way that computers and people can work collaboratively. In this sense, the RDF model provides such a definition by linking and representing resources and descriptions through defined schemes and vocabularies. However, much of the information able to be represented is contained within plain text, which results in an unfeasible task by humans to annotate large scale data sources such as the Web. Therefore, this paper presents a strategy for the extraction and representation of RDF statements from text. The idea is to provide an architecture that receives sentences and returns triples with elements linked to resources and vocabularies of the Semantic Web. The results demonstrate the feasibility of representing RDF statements from text through an implementation following the proposed strategy.
more …
By
ReyesNava, A.; CruzReyes, H.; Alejo, R.; RendónLara, E.; FloresFuentes, A. A.; GrandaGutiérrez, E. E.
Show all (6)
Performance of deep learning neural networks to classify class imbalanced geneexpression microarrays datasets is studied in this work. The low number of samples and high dimensionality of this type of datasets represent a challenging situation. Three sampling methods which have shown favorable results to deal with the class imbalance problem were used, namely: Random OverSampling (ROS), Random UnderSampling (RUS) and Synthetic Minority Oversampling Technique (SMOTE). Moreover, artificial noise and greater class imbalance were included in the datasets in order to analyze these situations in the context of classification of geneexpression microarrays datasets. Results show that the noise or separability of the dataset is more determinant than its dimensionality in the classifier performance.
more …
By
Rodríguez Sánchez, Alberto; Ponsich, Antonin; Jaimes, Antonio López; Martínez, Saúl Zapotecas
Show all (4)
Recent Multiobjective Optimization (MO) algorithms such as MOEA/D or NSGAIII make use of an uniformly scattered set of reference points indicating search directions in the objective space in order to achieve diversity. Apart from the mixturedesign based techniques such as the simplex lattice, the mixturedesign based techniques, there exists the Uniform Design (DU) approach, which is based on based on the minimization of a discrepancy metric, which measures how well equidistributed the points are in a sample space. In this work, this minimization problem is tackled through the
$$L_2$$
discrepancy function and solved with a parallel heuristic based on several Tabu Searches, distributed over multiple processors. The computational burden does not allow us to perform many executions but the solution technique is able to produce nearly Uniform Designs. These point sets were used to solve some classical MO test problems with two different algorithms, MOEA/D and NSGAIII. The computational experiments proves that, when the dimension increases, the algorithms working with a set generated by Uniform Design significantly outperform their counterpart working with other stateoftheart strategies, such as the simplex lattice or two layer designs.
more …
By
RamirezVelarde, Raul; HervertEscobar, Laura; HernandezGress, Neil
Organizations are turning to predictive analytics to help solve difficult problems and uncover new opportunities. Nowadays, the processes are saturated in data, which must be used properly to generate the necessary key information in the decision making process. Although there are several useful techniques to process and analyze data, the main value starts with the treatment of key factors. In this way, a Predictive Factor Variance Association (PFVA) is proposed to solve a multiclass classification problem. The methodology combines wellknown machine learning techniques along with linear algebra and statistical models to provide the probability that a particular sample belongs to a class or not. It can also give predictions based on regression for quantitative dependent variables and carryout clustering of samples. The main contribution of this research is its robustness to execute different processes simultaneously without fail as well as the accuracy of the results.
more …
By
SolorioFernández, Saúl; MartínezTrinidad, José Fco.; CarrascoOchoa, Jesús Ariel
Feature Selection for supervised classification plays a fundamental role in pattern recognition, data mining, machine learning, among other areas. However, most supervised feature selection methods have been designed for handling exclusively numerical or nonnumerical data; so, in practical problems of fields such as medicine, economy, business, and social sciences, where the objects of study are usually described by both numerical and nonnumerical features (mixed data), traditional supervised feature selection methods cannot be directly applied. This paper introduces a supervised filter feature selection method for mixed data based on the spectral gap score and a new kernel capable of modeling the data structure in a supervised way. To demonstrate the effectiveness of the proposed method, we conducted some experiments on public realworld mixed datasets.
more …
By
Sidorov, Grigori
After building the vector space model, we can represent and compare any type of objects of our study. Now we can discuss the question whether we can improve the vector space we have built. The importance of this question is related to the fact that the vector space model can have thousands of features, and possibly many of these features are redundant. Is there any way to get rid of the features that are not that important? Latent Semantic Analysis allows constructing new vector space model with smaller number of dimensions.
more …
By
Ita Luna, Guillermo; LópezRamírez, Cristina
We recognize the wheel graphs with different kinds of centers or axle faces as the basic pattern forming a planar graph. We focus on the analysis of the vertexcoloring for these graphic patterns, and identify cases for the 3 or 4 colorability of the wheels. We also consider different compositions among wheels and analyze its colorability process.
If a valid 3coloring exists for the union of wheels G, then our proposal determines the constraints that a set of symbolic variables must hold. These constraints are expressed by a conjunctive normal form
$$F_G$$
. We show that the satisfiability of
$$F_G$$
implies the existence of a valid 3coloring for G. Otherwise, it is necessary to use 4 colors in order to properly color G. The revision of the satisfiability of
$$F_G$$
can be done in polynomial time by applying unit resolution and general properties from equalities and inequalities between symbolic variables.
more …
By
Valdez, S. Ivvan; TrujilloRomero, Felipe
The Iterative Closest Point Algorithm (ICP) is a widely used method in computer science and robotics, used for minimizing a distance metric between two set of points. Common applications of the ICP are object localization and position estimation. In this work, we introduce a parallel version of the ICP which significantly reduces the computational time, by performing fewer operations while maintaining a simple and highly parallelizable algorithm. Our proposal is based on the naive computation of closest pairs of points in two different sets, instead of comparing all possible pairs we approximate the closest pairs of points by means of searching in a plausible subset. The experiments are performed on a sample from the Stanford 3D Scanning Repository, used for the 3D cloud of points registration. For these case studies, the error, as well as the solution, are exactly the same than using the exact algorithm.
more …
By
FranciscoValencia, Iván; MarcialRomero, José Raymundo; ValdovinosRosas, Rosa María
Monte Carlo Tree Search (MCTS) is the most used method in General Game Playing, area of the Artificial Intelligence, whose main goal is to develop agents capable of play any board game without preview knowledge. MCTS requires a tree which represents the states and moves of the board game which is visited and expanded using an iterations method. In order to visit the tree, MCTS requires a selection policy which determines which node is visited in each level. Nowdays, Upper Confidence Bound (UCB), is the most popular policy in MCTS due to its simplicity and efficiency. This policy was propose for the MultiArmed Bandit Problem (MABP) which consists in set of slot machines each of which has a certain probability of give a reward. The goal is to maximize the accumulative reward that is obtained when a machine is played in a series of rounds. Other policy proposed for MCTS is Upper Confidence Bound
$$_{\sqrt{.}}$$
(UCB
$$_{\sqrt{.}}$$
) whose goal is to identify the machine with the highest probability to give a reward. This paper shows a comparative between five modifications of UCB and one of UCB
$$_{\sqrt{.}}$$
, this comparative has the goal of finding a policy which be able to identify the optimal machine as quickly as possible, this goal in MCTS is equals to identify the node with the highest probability to leading to a victory. The results show that some policies find the optimal machine before UCB, however, with 10,000 rounds UCB is the policy who plays the optimal machine more often.
more …
By
Sidorov, Grigori
We have conducted various experiments [93] in order to test the usefulness of the concept of syntactic ngrams. Essentially, we consider the task of authorship attribution, i.e., there are texts for which the authors are known and a text for which we have to determine the author (among the considered authors only). In our case, we use a corpus composed of texts written by three different authors.
more …
By
MenesesViveros, Amilcar; ParedesLópez, Mireya; Gitler, Isidoro
In last years the use of multicore processors has been increased. This tendency to develop processors with several cores obeys to look for better performance in parallel programs with a lower consumption of energy. Currently, the analysis of performance of speedup and energy consumption has taken a key role for applications executed in multicore systems. For this reason, it is important to analyze the performance based on new characteristics of modern processors, such as Intel’s turbo boost technology. This technology allows to increase the frequency of Intel multicore processors. In this work, we present an extension of Amdahl’s law to analyze the performance of parallel programs running in multicore processors with Intel turbo boost technology. We conclude that for cases when the sequential portion of a program is small, it is possible to overcome the upper limit of the traditional Amdahl’s law. Furthermore, we show that for parallel programs running with turbo boost the performance is better compare to programs running in processors that does not have this technology on.
more …
By
RosasRomero, Roberto; LopezRincon, Omar; Starostenko, Oleg
The alpha matte is a twodimensional map that is used to combine two images, one containing a foreground and the other containing a background. Alpha matte extraction is performed on greenscreen images and requires user interaction to tune parameters in different preprocessing and postprocessing stages to refine an alpha matte. This paper tackles the problem of fully automatic extraction of the foreground on greenscreen images with extraction of the corresponding alpha matte. The method is based on a multilayer perceptron that assigns an alpha value, from a discrete set of ten alpha values, to each patch on a greenscreen image. The approach for assigning an alpha value to an image patch is based on a set of features that enhance discrimination between foreground and background. The classifier is trained to learn to separate foreground objects from greenscreen backgrounds as well as to generate the corresponding alpha matte map required for subsequent digital compositing. To test how the proposed approach handles alpha matte extraction under unsuitable conditions, a 64image dataset was generated. The main contribution is that our method overcomes two challenges publicly posed within a dataset of greenscreen image sequences, donated by Hollywood Camera Work LLC. Tests with this dataset generate highquality visual results for those two cases. These results are confirmed by comparing the proposed fully automatic alpha matte extraction with that based on the use of Adobe After Effects Creative Cloud, an application which heavily depends on user interaction.
more …
By
Aceves Gutiérrez, Luis Carlos; MartínGutiérrez, Jorge; Rio Guerra, Marta Sylvia
This paper presents fourteen useful guidelines to evaluate and determine how much does a digital interface in a public facility provide user experiences for citizens. In addition, it describes the partial outcomes of applying these guidelines to three urban interfaces in the city of Monterrey, Mexico.
An ethnographic study was conducted to obtain preliminaries results where the user experience was assessed in these interfaces when applying the proper guidelines. The goal is to validate guidelines with global assessment methods to establish a protocol of design for urban interfaces; in other words, an application that provides the information required to design urban interfaces from point of view of UX. This extended ethnographic study will be implemented in different cities around the world and urban interfaces.
more …
By
Bello, P.; Ita, G.
Let K be a Knowledge Base (KB) and let
$$\phi $$
be a query, the entailment problem
$$K \models \phi $$
is crucial for solving the Belief Revision problem. The Belief Revision problem consists in incorporate new beliefs into knowledge base already established, changing as little as possible the original beliefs and maintaining consistency of the KB. A widely accepted framework for reasoning through intelligent systems is the knowledgebased system approach. The general idea is to keep the knowledge in some representative language with a welldefined connotation, particularly it we will be used prepositional logic for modelling the knowledge base and the new information.
This article shows that the use of falsifying patterns for expressing clauses help to determine whether an conjunctive normal form (CNF) is inferred from another CNF, and therefore, it allows us to construct an algorithm for belief revision between CNF’s.
Our algorithm applies a depth first search in order to obtain an effective process for the belief revision between the conjuntive forms K and
$$\phi $$
.
more …
By
RechyRamirez, Ericka Janet; MarinHernandez, Antonio; RiosFigueroa, Homero Vladimir
1 Citations
Health conditions might cause muscle weakness and immobility in some body parts; hence, physiotherapy exercises play a key role in the rehabilitation. To improve the engagement during the rehabilitation process, we therefore propose a human–computer interface (serious game) in which five wrist movements (extension, flexion, pronation, supination and neutral) are detected via two commercial sensors (Leap motion controller and Myo armband). Leap motion provides data regarding positions of user’s finger phalanges through two infrared cameras, while Myo armband facilitates electromyography signal and inertial motion of user’s arm through its electrodes and inertial measurement unit. The main aim of this study is to explore the performance of these sensors on wrist movement recognition in terms of accuracy, sensitivity and specificity. Eight healthy participants played 5 times a proposed game with each sensor in one session. Both sensors reported over 85% average recognition accuracy in the five wrist movements. Based on t test and Wilcoxon signedrank test, early results show that there were significant differences between Leap motion controller and Myo armband recognitions in terms of average sensitivities on extension (
$$p = 0.0356$$
), flexion (
$$p = 0.0356$$
) and pronation (
$$p = 0.0440$$
) movements, and average specificities on extension (
$$p = 0.0276$$
) and pronation (
$$p = 0.0249$$
) movements.
more …
By
Kalashnykova, Nataliya; Avramenko, Viktor V.; Kalashnikov, Viacheslav
The paper develops an algorithm based on derivative disproportion functions (DDF) for modeling a cryptosystem for transmitting and receiving devices. The transmitted symbols are encoded with the aid of sums of at least two of those functions weighted with random coefficients. Some important properties of the derivative disproportion functions are also discussed. Numerical experiments demonstrate that the algorithm is quite reliable and robust.
more …
By
HerreraAlcántara, Oscar; GonzálezMendoza, Miguel
The article “Inverse formulas of parameterized orthogonal wavelets”, written by Oscar HerreraAlcántara and Miguel GonzálezMendoza, was originally published electronically on the publisher’s internet portal (currently SpringerLink) on 25 January 2018 without open access. The original article has been corrected.
more …
By
Sidorov, Grigori
In this and the following chapters, we present two ideas related to the nonlinear construction of ngrams. Recall that the nonlinear construction consists in taking the elements which form ngrams in a different order than the surface (textual) representation, i.e., in a different way than words (lemmas, POS tags, etc.) appear in a text. Here we consider filtered ngrams, i.e., we filter the elements before we start construction of ngrams.
more …
By
MarinCastro, Heidy M.; HernandezResendiz, Jaciel D.; EscalanteBalderas, Hugo J.; Pellegrin, Luis; TelloLeal, Edgar
Show all (5)
Image annotation is the task of assigning keywords or identifiers to images, holistically or in specific regions. These keywords serve as descriptors of highlevel semantics to facilitate retrieval and organization of visual information. It plays an important role in contentbased image understanding, as well as in areas such as object recognition in robotics, contentbased image searching and knowledge extraction. Automatic image annotation is usually approached by means of supervised classification, where a set of previously annotated images is required to train a learning algorithm that later predicts the labels for new images. This paper proposes a novel ensemble classifier for the supervised image annotation task inspired in chain classifiers. In the proposed approach a chain of individual classifiers is build, where each classifier is trained by using a different modality. In addition, the input space of models in the chain is augmented with the output of the preceding model in the sequence. Each model in the chain deals with the same classification problem, making the proposed method an ensemble model build from multimodal data. To the best of our knowledge, chain classifiers have not been used in this particular setting. Experimental results in a challenging image collection show that the proposed method is able to obtain an f −value superior to 0.5, outperforming related work.
more …
By
Beltrán, Jessica; Navarro, René; Chávez, Edgar; Favela, Jesús; SotoMendoza, Valeria; Ibarra, Catalina
Show all (6)
1 Citations
Frequently, people with dementia exhibit abnormal behaviors that may cause selfinjury or burden their caregivers. Some audible manifestations of these problematic behaviors are of vocal nature (e.g., shouting, mumbling, or cursing), others are environmental sounds (e.g., tapping or slamming). The timely detection of these behaviors could enact nonpharmacological interventions which in turn can assist caregivers or prevent escalation of the disruption with other fellow residents in nursing homes. We conducted a field study in a geriatric residence to gather naturalistic data. With the participation of five residents for 203 h of observation and of the 242 incidents of problematic behaviors were registered, 85% of them had a distinctive auditory manifestation. We used a combination of standard speech detection techniques, along with a novel environmental sound recognition methodology based on the entropy of the signal. We conducted experiments using realistic data, i.e., audio immersed in natural background noise. Based on classification results with F1 score above 87%, we conclude that audible cues can be used to enact nonpharmacological interventions aimed at reducing problematic behaviors, or mitigating their negative impact.
more …
By
SánchezAdame, Luis Martín; Mendoza, Sonia; Meneses Viveros, Amilcar; Rodríguez, José
Show all (4)
Thanks to the hyperconnected world in which we live, we are surrounded by devices. Many of these devices can communicate with each other, and even they can support the same application so that the user can have multiple forms of interaction. However, developers should be careful when considering the control level left to users, since the applications may become unusable. Whether the system or user decides the distribution of the available devices, Graphical User Interface (GUI) consistency must always be preserved. Consistency not only provides users with a robust framework in similar contexts but is an essential learning element and a lever to ensure the GUI efficient usage. This paper proposes a set of consistency guidelines that serve as a means for the construction of multidevice applications. As a case study, DistroPaint was evaluated by experts who identified consistency violations and assessed their severity.
more …
By
GaxiolaTirado, J. A.; RodríguezUgarte, M.; Iáñez, E.; Ortiz, M.; Gutiérrez, D.; Azorín, J. M.
Show all (6)
Transcranial direct current stimulation (tDCS) is a noninvasive technique for brain stimulation capable of modulating brain excitability. Although beneficial effects of tDCS have been shown, the underlying brain mechanisms have not been described. In the present study, we aim to investigate the effects of tDCS on EEGbased functional connectivity, through a partial directed coherence (PDC) analysis, which is a frequencydomain metric that provides information about directionality in the interaction between signals recorded at different channels. The tDCS montage used in our study, was focused on the lower limbs and it was composed of two anodes and one cathode. A singleblind study was carried out, where eight healthy subjects were randomly separated into two groups: sham and active tDCS. Results showed that, for the active tDCS group, the central EEG electrodes Cz, C3 and C4 turned out to be highly connected within alpha and beta frequency bands. On the contrary, the sham group presented a tendency to be more random at its functional connections.
more …
By
Quintero, Rolando; TorresRuiz, Miguel; MenchacaMendez, Rolando; MorenoArmendariz, Marco A.; Guzman, Giovanni; MorenoIbarra, Marco
Show all (6)
This paper presents the DISC approach, which is a novel method to assess the conceptual distance between concepts within an ontology. DISC is graph based in the sense that the whole topology of the ontology is considered when computing the weight of the relationships between concepts. The methodology is composed of two main steps. First, in order to take advantage of previous knowledge, an expert of the ontology domain assigns initial weight values to each of the relations in the ontology. Then, an automatic method for computing the conceptual relations refines the weights assigned to each relation until reaching a stable state. We introduce a metric called generality that is defined in order to evaluate the accessibility of each concept, considering the ontology like a strongly connected graph. Unlike most previous approaches, the DISC algorithm computes similarity between concepts in ontologies that are not necessarily represented in a hierarchical or taxonomic structure. So, DISC is capable of incorporating a wide variety of relationships between concepts such as meronymy, antonymy, functionality and causality.
more …
By
Camiña, José Benito; MedinaPérez, Miguel Angel; Monroy, Raúl; LoyolaGonzález, Octavio; Villanueva, Luis Angel Pereyra; Gurrola, Luis Carlos González
Show all (6)
1 Citations
Dependence on personal computers has required the development of security mechanisms to protect the information stored in these devices. There have been different approaches to profile user behavior to protect information from a masquerade attack; one such recent approach is based on user fileaccess patterns. In this paper, we propose a novel classification ensemble for file accessbased masquerade detection. We have successfully validated the hypothesis that a oneclass classification approach to file accessbased masquerade detection outperforms a multiclass one. In particular, our proposed oneclass classifier significantly outperforms several stateoftheart multiclass classifiers. Our results indicate that oneclass classification attains better classification results, even when unknown attacks arise. Additionally, we introduce three new repositories of datasets for the identification of the three main types of attacks reported in the literature, where each training dataset contains no object belonging to the type of attack to be identified. These repositories can be used for testing future classifiers, simulating attacks carried out in a real scenario.
more …
By
SánchezSánchez, Christian; Sheremetov, Leonid B.
Despite increasing availability of Web Services (WS), their automatic processing (classification, grouping or composition) slows down because of the difficulty to read the WSDL service descriptions without related technical knowledge. Categorizing services for automatic service discovery and composition has become a challenging problem. The paper argues that ngram representation of the data extracted from the different sections of the WSDL description (types, messages and operations) along with the weighing scheme can benefit the classification of services. Experiments are carried out with three different classifiers over available collections of WS descriptions. It is shown that such representations as word bigrams or letter trigrams extracted from WSDL Operations and Types service description features with TFIDF as ngram weighting scheme, can improve automatic WS classification.
more …
By
Sidorov, Grigori
As we described in the previous chapters, mainstream of the modern computational linguistics is based on application of machine learning methods. We represent our task as a classification task, represent our objects formally using features and their values (constructing vector space model), and then apply wellknown classification algorithms. In this pipeline, the crucial question is how to select the features. For example, we can use as features words or ngrams of words (sequences of words) or sequences of characters (character ngrams), etc. An interesting question arises: Can we use syllables as features? It is very rarely done in computational linguistics, but there is certain linguistic reality behind syllables. This chapter explores this possibility for the authorship attribution task; it follows our research paper [99]. Note that syllables are somewhat similar to character ngrams in the sense that they are composed of several characters (being not too long).
more …
By
Obeso, A. Montoya; BenoisPineau, J.; Vázquez, M. S. García; Acosta, A. A. Ramírez
Show all (4)
The automatic description of digital multimedia content was mainly developed for classification tasks, retrieval systems and massive ordering of data. Preservation of cultural heritage is a field of high importance of application of these methods. We address classification problem in cultural heritage such as classification of architectural styles in digital photographs of Mexican cultural heritage. In general, the selection of relevant content in the scene for training classification models makes the models more efficient in terms of accuracy and training time. Here we use a saliencydriven approach to predict visual attention in images and use it to train a Deep Convolutional Neural Network. Also, we present an analysis of the behavior of the models trained under the stateoftheart image cropping and the saliency maps. To train invariant models to rotations, data augmentation of training set is required, which posses problems of filling normalization of crops, we study were different padding techniques and we find an optimal solution. The results are compared with the stateoftheart in terms of accuracy and training time. Furthermore, we are studying saliency cropping in training and generalization for another classical task such as weak labeling of massive collections of images containing objects of interest. Here the experiments are conducted on a large subset of ImageNet database. This work is an extension of preliminary research in terms of image padding methods and generalization on large scale generic database.
more …
By
SerranoRubio, A. Alejandra; MenesesViveros, Amilcar; MoralesLuna, Guillermo B.; ParedesLópez, Mireya
Show all (4)
A parallel system to solve complex computational problems involve multiple instruction, simultaneous flows, communication structures, synchronisation and competition conditions between processes, as well as mapping and balance of workload in each processing unit. The algorithm design and the facilities of processing units will affect the costperformance ratio of any algorithm. We propose a generic methodology to capture the main characteristics of parallel algorithm design methodologies, and to add the experience of expert programmers through pattern languages. Robust design considering the relations between architectures and programs is a crucial item to implement highquality parallel algorithms. We aim for a methodology to exploit algorithmic concurrencies and to establish optimal process allocation into processing units, exploring the lowest implementation details. Some basic examples are described, such as the kmeans algorithm, to illustrate and to show the effectiveness of our methodology. Our proposal identifies essential design patterns to find models of Data Mining algorithms with string selfadaptive mechanisms for homogeneous and heterogeneous parallel architectures.
more …
By
Kalashnykova, Nataliya; LealCoronado, Mariel A.; GarcíaMartínez, Arturo; Kalashnikov, Viacheslav
Show all (4)
We study the effects of merging two separate markets each originally monopolized by a producer into a globalized duopoly market. We consider a linear inverse demand with cap price and quadratic cost functions. After globalization, we find the consistent conjectural variations equilibrium (CCVE) of the duopoly game. Unlike in the Cournot equilibrium, a complete symmetry (identical cost functions parameters of both firms) does not imply the strongest coincident profit degradation. For the situation where both agents are lowmarginal cost firms, we find that the company with a technical advantage over her rival has a better ratio of the current and previous profits. Moreover, as the rival becomes ever weaker, that is, as the slope of the rival’s marginal cost function increases, the profit ratio improves.
more …
By
León, José Alfredo Sánchez
Binomials and multinomies are mathematical functions that do appear in many fields like linear algebra, calculus, statistics and probability, among others. Complete binomial and multinomial construction can be a hard task; there exist some mathematical formulas that can be deployed to calculate binomial and multinomial coefficients, in order to make it quicker. The purpose of this document is the development of an alternative method in order to carry out the calculation of binomial and multinomial coefficient; here are raised three analytic formulas that yield those coefficients for each term by means of summations series. Throughout this document firstly it is exposed the deduction of the two formulas to calculate binomial coefficients, afterwards this result is extended alongside the binomial theorem for the n terms of a multinomial to code a formula that can be used for multinomies. Finally an exemplary case of study is bestowed in order to illustrate how to deploy those formulas.
more …
By
MillánHernández, Christian Eduardo; GarcíaHernández, René Arnulfo; Ledeneva, Yulia; HernándezCastañeda, Ángel
Show all (4)
Lookalike and Soundalike drug names are related to medication errors where doctors, nurses, and pharmacists prescribe and administer the wrong medication. Bisim similarity is reported as the best orthographic measure to identifying confusable drug names, but it lacks from a similarity scale between the bigrams of a drug name. In this paper, we propose a SoftBisim similarity measure that extends to the Bisim to soften the comparison scale between the Bigrams of a drug name for improving the detection of confusable drug names. In the experimentation, SoftBisim outperforms others 17 similarity measures for 396,900 pairs of drug names. In addition, the average of four measures is outperformed when Bisim is replaced by SoftBisim similarity.
more …
By
Collazos, C. A.; Sanz, Fredy; Monroy, I. A.; MaldonadoFranco, Adriana; DelaHozFranco, Emiro; MeléndezPertuz, Farid; Mora, César
Show all (7)
This article shows the results obtained in the application of State Estimation to a case of fault detection and diagnosis in an osmotic dehydration process. The disturbances that affect the measurements on the system are unknown but it is known that they are bounded. The tools used for the development of this work were the EMV (Extended Mean Value) and SIVIA (Set Inversion Via Interval Analysis) algorithms, which allow obtaining the necessary boundaries for the fault detection and diagnosis algorithm. The paper presents simulations and validations.
more …
